Qwen 3 with Ollama and VS Code: local coding setup

This tutorial sets up Qwen as a local coding assistant in VS Code through Ollama and the Qwen Code Companion extension. The presenter installs Ollama on Windows, downloads a model through the terminal, and checks its responses in Ollama before configuring the editor. Although the title says Qwen 3, the model selection discussion refers to Qwen 3.6 and a download of about 24 GB. The transcript does not clearly establish the exact model tag selected.

The VS Code setup uses Qwen Code Companion, which the presenter describes as being in preview. In its custom provider settings, the walkthrough selects OpenAI compatibility and points the connection at localhost:11434/v1. The presenter enters a placeholder API key and stresses copying the exact model name from Ollama. These are the concrete connection steps for readers who want to run models locally through this editor workflow.

The coding demonstration asks Qwen to generate a simple HTML page, opens the result, and then requests dark styling and a responsive layout. The final page includes light and dark mode options. This is a small example rather than a broad assessment of coding accuracy. In the earlier chat, the model reports that it cannot browse the live internet or execute code on the device; the presenter distinguishes that chat behavior from an agent setup.